Analysing content of Paris climate pledges with computational linguistics

Parties to the Paris Agreement submit their climate action plans, known as nationally determined contributions (NDCs), which outline mitigation targets and strategies to achieve them. While existing research has focused on assessing the mitigation targets, there is a wealth of broader textual content within the documents that has received little attention. Using natural language processing to systematically analyse the full textual content of all NDCs, we identify 21 topics that form seven thematic groups: development, implementation and planning, mitigation targets, policies and technologies, climate change impacts, agriculture and ecosystems, and stakeholders. We also examine how attention to specific topics has evolved over time and across parties. We find that high-income countries, typically shouldering greater historical responsibility for emissions, tend to focus on mitigation targets but provide limited detail on concrete policies being implemented. In contrast, developing countries often frame their NDCs within broader visions of sustainable development, balancing mitigation with adaptation and competing development goals. Establishing a standardized, transparent NDC format could enhance comparability of the plans of parties and assessment of how mitigation targets will be achieved while balancing trade-offs and co-benefits with other sustainable development goals.